This kind of came up in another thread and starting me thinking that I have rarely departed from Source Alpha .. compositing in Vegas is still a bit of mystery to me. So hence my question. What is Source Alpha and why should I care?
You should care, because it gives you the posibility to combine two or more images in different ways... in other programs this is also known as "blending modes". Source alpha is the default where no changes are made to the transparency of your video event
From vegas help file:
Alpha is a fourth channel that determines how transparency is handled in an image file. The RGB channels are blended to determine each pixel's color, and the corresponding alpha channel determines each pixel's transparency. The alpha channel can have up to 256 shades of gray: 0 represents a transparent pixel, 255 represents an opaque pixel, and intermediate values are semitransparent.
So you apply the Alpha channel to the entire track. Source Alpha is the default .. that's normal .. no worries .. no transparency .. but now the fun begins .. There's a whole menu list of modes .. what the heck is "Difference Squared" and how would anyone every know how to use it? Yes. I can read the Vegas help for Difference Squared but it doesn't really explain much. It's hardly intuitive. Is this stuff best left to the Compositing Guru's ?
what the heck is "Difference Squared" and how would anyone every know how to use it?
Presumably, it is the square (as in x²) of the difference between what is on the current track and what is on the one below it. I used it, for example, when comparing the effect of a LUT on one track and the corresponding matrix on another. Since both, the difference and the difference squared, gave me a solid black, it was a fairly good confirmation the matrix had the same effect as the LUT. I discussed that in this thread.
Anyway, Chapter 7 of Digital Compositing for Film and Video (Third Edition), by Steve Wright discusses the various blending modes. It is not Vegas specific, but it is a must-read book for anyone who does any kind of video compositing. Read it, and then you’ll be the Compositing Guru. 😉
I can read the Vegas help for Difference Squared but it doesn't really explain much
The way I think of it, computers are great at doing mathematical manipulations of data sets. You take a data set such as a video or audio file and throw a number of known mathematical manipulations at it and sometimes something such as convolving your video data with a Gaussian function, playing with some of the parameters ends up with a useful blurring effect. Other easily applied mathematical functions such as difference squared may end up producing results that could produce visually useful outcomes. It's a bit like the analogue days of pouring a bit of your aunt's raspberry vinegar (in this case, the difference squared analysis) into the developing liquid to see if it produced a unique effect that was useful.
So, as a a simple example I might use the difference squared option as one ingredient when conjuring up a Ghost of Christmas Past in Scrooge's bedroom.
Xberk, use the mode that you like artistically and dispense with the math. Unless you're like me who sickens myself on this stuff.
Just read a little and experiment a lot...
http://www.digitalproducer.com/article/Understanding-Video-Compositing-1652441
http://nofilmschool.com/2013/05/premiere-pro-tutorial-adobe-blend-modes
And unleash your artistic vein... try it with light flares, grain, duplicating your tracks, special fx (smoke, fire, etc), try with 2, 3 or more different video tracks
Ok. This came up today. I had some interview footage that was very flat and underexposed. I did some CC with Color Curves and got it looking fairly decent rather quickly. Now I've gone one step further using composting. This little gem came from Rick Wise on Creative Cow .. He posted this back in 2009. I was not really looking for this specifically. I was just Googling "sony vegas difference squared" and up popped something I could use immediately.
"... I find it useful to copy the rendered/color-corrected timeline to the line above; turn the top line 100% b&w, increase sharpness to "medium"; reduce the top to 25-30% level; the result then is a quite desaturated image, so I increase saturation on the bottom level by eye using the saturation slider in Color Corrector. The result controls over-exuberant highlights and adds a slightly more sharp and more nuanced picture than the original. The change is slight, but worth the effort in certain cases."
The top image is the original, uncorrected. The middle is with Color Curves alone. The bottom is composited per Rick Wise's method. I dropped the Level on the top track to 60% and it gave a better result. I like it.
Good job, xberk. 😎